Floating ground, PSU woes

[Aeternam]

Hello gang,

Utter newbie here so please be gentle. I recently got the electronics bug and after a lot of reading (and wondering at how much stuff I forgot since attending my last physics class) I went out and bought my first multimeter and a multichannel power supply. I'm working on my first project: a constant DC load, heavily inspired by Dave's. I got it wired up on my breadboard and it seems to be working reasonably well -- but I have a stupid question.

The controlling circuit uses +12V so I have that one wired to channel 1 on my PSU. Channel 2 is wired so that it loops through the MOSFET and the current sense resistor. So far so good. Funny thing is, this only works well when I use a COMMON ground (ie. when I connect the negative lead of channel 2 to the negative lead of channel one, on the breadboard ofc.) Is this to be expected? Does this have to do with "floating ground"?

Furthermore: a bit of investigation has shown that connecting the + line of channel 1 to the - lead of channel 2 shows a voltage difference even if the supply is not supplying current (soft-turned-off). I find this hard to understand. Am I missing something? Or is my PSU faulty?

PSU: http://www.reichelt.de/Labornetzgeraete/KA3305P/3/index.html?ACTION=3&GROUPID=4952&ARTICLE=148161&OFFSET=500&WKID=0&SID=15VkICK6wQATcAAK0BYt0ebe7c4116cbede0afd6a6d557a535014&LANGUAGE=EN

Multimeter is a 87V so I have good confidence that the readings are correct.

Thanks!

[macboy]

The two channels of the power supply are not internally connected. They are floating with respect to each other. If you wanted to, you could connect them in series (negative of ch 2 to positive of ch 1 or vice-versa) to get higher voltage. There is no restriction of how you connect the two of them. Think of them like two separate adjustable batteries. Also note that both channels are floating with respect to earth/ground. You may (if you choose to) connect any reasonable point in your circuit to one of the ground terminals of the power supply, to provide a ground reference, but this is not strictly necessary. In fact, if you are going to use an oscilloscope to probe signals in the circuit, you usually want to leave the circuit floating, to prevent ground loops (through the additional ground connection of the scope probe).

That said, you must connect both the positive and negative terminal of each channel to something in your circuit. From your description it sounds like you did not connect the negative/black side of channel 1 (12 V) power supply.  In the basic Dave's DC load, the circuit's op-amps power and the input (load) power are both connected to a common circuit "ground" (not to be confused with real "earth" ground, a circuit "ground" is just the "zero volt" reference for the circuit... it is not necessarily connected to a real earth/ground). So the circuit worked when you connected both negative terminals together at the breadboard.

Speaking of breadboards, be careful how much current you put through it. You can and should do the high current connections (the MOSFET and current sense resistor) off the breadboard.

[stmdude]

Think of them like two separate adjustable batteries.

Yup, stealing that description for the next time people at work ask me why their PSUs have three connections (+,-,Earth).

[Aeternam]

In the basic Dave's DC load, the circuit's op-amps power and the input (load) power are both connected to a common circuit "ground"

Yes I see that now. I was working under the assumption that the controlling circuit and the controlled circuit were completely independent. Clearly this is not the case because of the opamp's feedback loop which connects the two. Should have been obvious from the start

Thanks for clearing that up for me